Posted: June 20, 2014
Low-density parity-check (LDPC) codes are used in communications technologies to successfully send a message over a noisy channel. These codes come much closer to the theoretical maximum transmission rate than other methods. Unfortunately the conventional (logarithm likelihood ratio or LLR) message scheme of LDPC decoders requires complex algorithms involving time, power, and memory intensive hyperbolic tangent functions.
Researchers at the University of Alberta have developed two hybrid message decoding algorithms that reduce the complexity of the computations required to decode a message. One decoding algorithm uses a hybrid between likelihood ratio (LR) and probability difference (PD) metrics whereas the other uses a hybrid between logarithm likelihood ratio (LLR) and logarithm probability difference (LPD) metrics. Variable and check node processing within the LDPC codes using these new algorithms can be done with simple additions, subtractions, multiplications and divisions. These reductions in computational complexity enable a faster decoder chip that requires less power and memory.
- Reduced computational complexity (uses only 4 elementary operations).
- Decreased decoding time.
- Reduction in power and memory requirements.
Companies that produce decoder chips for LDPC codes (most chip manufacturers) would greatly benefit from this technology. Many current industry standards such as the DVB-S2 standard for satellite television, the IEEE 802.11n wireless LAN standard, IEEE 802.3an, and IEEE 802.16 for wireless metropolitan area networks utilize LDPC codes.